Effect Of Data Preprocessing on the Forecasting Accuracy of Solar Power Plant

Accurate forecasting is one of the bases of the successful functioning of renewable energy sources. The use of artificial intelligence methods makes it possible to achieve high forecasting accuracy. However, the forecast accuracy depends not only on the forecasting methods used, but also on the quality of the initial data. This paper described the most common metrics for evaluating forecasting accuracy and its interpretation. Also, the frequent defects of initial data were illustrated in this paper using the example of real solar power plant generation data. The linear regression model trained using machine learning methods was used as a forecasting model. The analysis of forecasting model results was performed using the evaluation metrics which were described. By the analysis of the forecasting model results the effects of cleaning and preparation of initial data on the forecasting accuracy of a solar power plant were shown.